Feasibility of Bluetooth Data as a Surrogate Analysis Measure of Traffic

Background The proliferation of portable electronic devices among consumers has created in recent times new opportunities for traffic data collection. Many of these devices contain short range Bluetooth radios in addition to other electronic equipment. The included Bluetooth radio on each device was intended to provide a low-power communications protocol to connect devices such as cell phones, headphones, music players, and more to each other. The presence of a unique identification number as part of the Bluetooth protocol on each device, that when activated can be discovered electronically, unintentionally creates anonymous probes in the traffic stream. This research explored possibilities of using Bluetooth technologies for various traffic data collection studies to expand the tools available to traffic engineers. Data Collection This study began with testing Bluetooth roadside data logger hardware configurations. Controllable variables included Bluetooth antenna selection and roadside placement options. Through the use of controlled conditions, detection areas for five antenna options were mapped, and their detection reliabilities were assessed. Other tests were conducted to assess the impacts of roadside antenna placement, vehicular speeds and in-vehicle source placement. This research then builds on the data collected about Bluetooth hardware performance metrics by investigating the feasibility of using Bluetooth data as a surrogate for traditional traffic engineering data for several traffic study applications. These studies included: urban corridor travel time monitoring, freeway travel time monitoring, origin-destination studies, and estimating turning movements at roundabouts. Each of these studies was parallel in nature to each other and showed how the same technology could be applied to different study objectives. Analysis The data collected during each of the studies provided valuable insight into Bluetooth technology. The hardware evaluations showed that a dipole antenna placed 6-12 feet from the edge of the roadway with at least 3 feet of elevation performed the best. The antenna power of the dipole could be changed to increase or reduce the coverage area as needed. The urban corridor study found that the Bluetooth data collection method provided similar results in a before-after analysis as GPS probe vehicles. The urban freeway corridor study found statistically significant differences in travel time data compared with permanent travel time sensor data provided by the regional traffic management center for seven of the eight freeway corridor segments tested. However, these differences were small and appeared not to be practically significantly different. The origin-destination study found no significant differences for either travel times or percentage or through trips between Bluetooth data collection and video re-identification of vehicles. Finally, the roundabout study showed that estimates of turning movement counts could be successfully accomplished, but in one case was significantly different than manual count data; additional research is needed to better understand the differences in roundabout turning movement counts. Conclusions The use of Bluetooth technology showed new possibilities for data collection. The data collected allowed for an automated process for identifying and re-identifying vehicles along a corridor. Traditional traffic study methodologies, such pairing of vehicular data or simply observing (counting) traffic flows, required many hours of labor intensive data collection that could be replicated with Bluetooth technology in a matter of minutes. Additionally, Bluetooth data sets opened up new potential analyses of the data. Such additional analyses included being able to separate frequent (repeat) travelers from occasional travelers along a corridor. While this technology was found to have enormous potential, it was not found to be completely stand-alone. The chief weakness of the technology was that it was found to sample around 5 percent of the available traffic. The implication of this was that Bluetooth data were not always available or sufficient in size for analysis. This could be a particular issue when one needs to delineate a day into small time frames. Furthermore, because of this unintentional use of Bluetooth technology, there was not any way to guarantee data to be available at the time periods needed. Also, in order to extrapolate volumetric data from the Bluetooth data, a secondary source was needed to assess a Bluetooth penetration rate. Thus the abandonment of current technologies and methodologies would undermine this data collection technique. A key assumption was that each Bluetooth source detected represented a separate independent vehicle. While this assumption could be violated with multiple discoverable Bluetooth devices in a single vehicle (e.g. a transit bus), this was not found to be an issue. Through this research it has been shown that the use of Bluetooth technology has earned its place in an engineer's toolbox